I think that what is normally being monitored in IT-systems is in fact insulation, rather than isolation. (I’m told that in some parts of the world the practice is to measure isolation impedance rather than insulation resistance, but then the measurement is normally given as a prospective mA current value rather than a kohm value).
If your insulation resistance is being measured by a dc monitoring signal then any errant dc present (eg: from a faulty s-m power supply) will add or subtract to the measured value, causing over- or under-sensitivity respectively. In the former case this could result in nuisance alarming, the latter in loss of safety margin.
Potentially relevant to your case is that in the Bender sales literature, (Bender are also manufacturers of IT-power systems), they report having found a particular problem with a number of makes of VDU appliances. Apparently a certain internal EMC fix - of which unfortunately I have no knowledge - used in some of these devices results in small dc current flow of the type described above, which then affects results of the IT insulation monitoring as described. If your Starkstrom IT-system uses a plain dc monitoring signal then this might be the source of your problem. (I believe Bender now use a dc signal which is pulsed at very low frequency, which then enables the monitor to detect the presence of any other dc signals and correct results accordingly).
p.s.: With regard to comments on IT-system monitoring above , insulation, overcurrent and temperature are all requirements of IEC-60364-7-710 so will now be the norm. (The relevant chapter in the IEE’s Guidance Note 7 (2003) supporting the ‘Wiring Regs’ is based upon this IEC Standard).
